Tracing Galaxy Formation with Stellar Halos II: Relating Substructure in Phase- and Abundance-Space to Accretion Histories

Draft version July 24, 2008

Preprint typeset using I^T^^X style eniulateapj v. 10/09/06

TRACING GALAXY FORMATION WITH STELLAR HALOS II:
RELATING SUBSTRUCTURE IN PHASE- AND ABUNDANCE-SPACE TO ACCRETION HISTORIES

00
O
O

(N

Oh
O

Kathryn v. Johnston ^ James S. Bullock^, Sanjib Sharma\ Andreea Font^, Brant E. Robertson^

N. Leitner^

Draft version July 24, 2008

AND Samuel

ABSTRACT

This paper explores the mapping between the observable properties of a stellar halo in phase- and
abundance-space and the parent galaxy's accretion history in terms of the characteristic epoch of
accretion and mass and orbits of progenitor objects. The study utilizes a suite of eleven stellar halo
models constructed within the context of a standard ACDM cosmology. The results demonstrate
that coordinate-space studies are sensitive to the recent (0-8 Gyears ago) merger histories of galaxies
(this timescale corresponds to the last few to tens of percent of mass accretion for a Milky- Way-type
galaxy). Specifically, the frequency, sky coverage and fraction of stars in substructures in the stellar
halo as a function of surface brightness are indicators of the importance of recent merging and of the
luminosity function of infalling dwarfs. The morphology of features serves as a guide to the orbital
distribution of those dwarfs. Constraints on the earlier merger history (> 8 Gyears ago) can be gleaned
from the abundance patterns in halo stars: within our models, dramatic differences in the dominant
epoch of accretion or luminosity function of progenitor objects leave clear signatures in the [a/Fe]
and [Fe/H] distributions of the stellar halo — halos dominated by very early accretion have higher
average [a/Fe] , while those dominated by high luminosity satellites have higher [Fe/H] . This intuition
can be applied to reconstruct much about the merger histories of nearby galaxies from current and
future data sets.

Subject headings: Galaxy: evolution — Galaxy: formation — Galaxy:halo — Galaxy: kinematics
and dynamics — galaxies: dwarf — galaxies: evolution — galaxies: formation
— galaxies: halos — galaxies: kinematics and dynamics — Local Group — dark
matter

en

o

00
O

>

X

1. INTRODUCTION

Phase- and abundance-space substructure in the stellar
distribution around a galaxy is commonly interpreted as
a natural outcome of the process of hierarchical struc-
ture formation, where large galaxies are built in part
from the accretion of dwarf galaxies. Numerous previ-
ous studies have looked at how to understand individual
debris features around galaxies i n terms of the proper-
ties of the progenitor dwarfs (e.g. Johnston 1998: Hclmi
fc White|[T999l [Johnston, Sackett,^ Bullock..2001. ,Law

et al.|2005[|Fardal et aL|2006||Warnick et al.|2008p . More
generally, we might ask: to what extent can the merger
history of a galaxy be reconstructed from its surrounding
substructure?; and what could you learn about the na-
ture of merger histories in our Universe by examining the
outskirts of a large sample of galaxies to very low surface
brightness? These questions have yet to be addressed
beyond using simple ana lytic estimates for the expected

scalings in tidal debris (Bullock et al.
Sackett, k Bullock 200ll

2001 Johnston,

^ Department of Astronomy, Columbia University, Pupin
Physics Laboratory, 550 West 120th Street, New York, NY 10027,
USA

^ Center for Cosmology, Department of Physics & Astronomy,
University of California, Irvine, CA 92697, USA; buUockCSuci.edu

^ Institute for Computational Cosmology, University of Durham,
South Road, Durham DHl 3LE, United Kingdom

"^ Kavli Institute for Cosmological Physics, Department of As-
tronomy and Astrophysics, University of Chicago, 933 East 56th
Street, Chicago, IL 60637

5 Enrico Fermi Institute, 5640 S. Ellis Ave. Chicago, IL 60637

^ Spitzer Fellow

Motivation for answering these questions in more depth
has been amply provided in the last decade by obser-
vations which have revealed abundant substructure in
the spatial distribution around both the Milky Way and
Andromeda galaxies, already seen at a few tens of kpc

from their centers, and dominant at lOOkpc (e.g. Ivezic

20021

et al.

Majewski et al. |2003||Belokurov et al |2006[ |Ibata

erar||2000| [Newberg et al.|[2002| |2003| ll-'erguson et al

2007 and see !6.1 for more examples). The ubiq-

uity of such substructure has become apparent because
of dramatic increases in the sky-coverage of halo surveys,
the number of tracers (and hence surface brightness to
which such surveys are sensitive) and the distances from
the parent galaxies which have been probed. In contrast,
only slightly more than a decade ago the classical picture
of stellar halos around galaxies was that the stars in them
were smoothly distributed in phase-space — a picture
informed by observations of the RR Lyrae and globular
cluster distribution around the Milky Way u sing samples
of a few dozen to hundreds of objects (e.g. Wetterer &
McGraw 1996"). In our own Galaxy, even halo stars in
the Solar neighborhood have been shown to be clumped
once their full phase-space coordinate s are known and
their orbital dis tributio n is considered ( Helmi, White, de
Zeeuw fc Zhao; 1999 M orrison et "ar]|2008p .

Satellite accretion is the preferred explanation for ap-
parent phase-space substructure. Debris from the de-
struction of a satellite will phase-mix along the progeni-
tor's orbit over time to fill the full volu me of the original

orbit in coordinate space (Johnstonll998), while becom-

ing locally colder at each point in velocity-space (in order

to satisfy Louiville's theorem — see Helmi fc White|1999
for a rigorous description). The small range in orbital
angular momenta and energy is largely conserved during
any gradual evolution of the parent ga laxy potential, al-
though the average val ues can evol ve (Penarrubia et al.'
[2006 J . (See also Knebe et al.|[2005| [Warn ick et al. .2008,
for a discussion in a more violent context.) Putting these
results for individual accretion events together leads to
a spectrum of lumps in phase-space, as has been demon-

strated in composite stvidics of halo distributions ( Helmi
& de Zecuw 2000: Bullock et al. 2001: Johnston, Sackett,

fc Bullo ck 2001: Helmi, White, & Springe ll2003| [Bullock
fc Johnston| |2005i |De Lucia fc Helmi | |2008|.

Similarly, substructure is now becoming apparent in
metallicity distributions and abundance patterns of halo
stars. So far, these studies have relied on using coor-
dinates in both phas e- and abundance-space to identify

these structures (e .g. 'Majew ski et al.|1996
[2006:'lbataet al. 12007

2004

2008

IHelmi et al

However, a

Navarro et al
[Gilbert et al.

bundance space has the advantage over
phase-space of preserving a complete memory of a star's
initial conditions which cannot be destroyed by subse-
quent mixing or scattering. In principle, "chemically tag-
ging" stars could lead to as sociations that are not appar

entin phase-space alone (Bland-Hawthorn fc Freeman
2003 1. Indeed, it has already been demonstrated that
the distinct merging time and mass scales of stellar halo,
satellite and dominant substructure progenitors means
that stars in each of these systems should have distinct

patterns in a-element/metallicity space (Robertson et al.
[2005, Font et al. 2006a b , 2008 ), which imphes that dis-
tinctions between lower-contrast systems could be pos-
sible in higher dimensions. These results also suggest
that abundance distributions in a stellar halo should re-
flect its accretion history (as has alread y been sho wn for
metall icity distributions: [Renda et al.[|2005 Font et al.

2006b I.

I'his paper takes a first step towards addressing how to
broadly relate substructures apparent today to a galaxy's
past by exploring how the observed properties of eleven
stellar halo models built entirely from accretion events
within the context of a ACDM universe reflect their ac-
cretion histories. Further work will build on the intuition
developed here to deflne statistical measures of substruc-
ture that are tuned to be sensitive to the epoch of ac-
cretion and mass and orbit type of progenitor satellites.
The modeling methods are described in §2. In §3 the
1515 simulated accretion events from all eleven model
halos are first analyzed individually in order to charac-
terize how the observed properties of debris (i.e. spatial
and velocity scales, morphology, surface brightness and
abundance patterns) are related to the intrinsic proper-
ties of the progenitor satellite (i.e. accretion time, lumi-
nosity and orbit). The intuition developed in §3 is then
applied in §4 to understanding how the observed proper-
ties of stellar halos (i.e. frequency, morphology, surface
brightness and stellar populations of non-uniform fea-
tures) arise from their unique formation histories. These
ideas are illustrated by contrasting our eleven "standard"
(i.e. built within a ACDM context) halos with four model
halos built from accretion histories that have been ar-
tificially constrained to be dominated by ancient/recent
and high/low luminosity events, as well as two more built
from events predominantly on radial/circular orbits. The

results are reviewed in §5, and subsequently applied to
"observations" of our model halos in order to demon-
strate in principle how such data might be interpreted.
In §6 we discuss applications of these ideas to both cur-
rent and future data sets. We summarize our conclusions
in §7.

2. MODELING METHODS

Our models were constructed to provide high-
resolution examples of what stellar halos around Milky-
Way-type galaxies built from accretion within the con-
text of a ACDM universe might look like. A concise
description is included below. (For full det ails of the
method and tests of our models see [Bu llock fc JohnstonI
2005 ; ^ Robertson eraI1[2005| [Font et al [ [2006a )

The modeling method can be broadly split into three
phases:

Phase I: Simulating the dark matter dynamical evolu-
tion — The time and mass of each accretion event
throughout a galaxy's history was first generated at
rand om from an extended Press-Schechter merger
tree ( Somerville fc Kolatt|[r999| ). The events were
assigned orbital eccentricities and binding energies
drawn from orbital distributions of satellites ob-
served in fully self-consistent cosmological simu-
lations of structure formation. Individual, high-
resolution N-body simulations were then run to
track the evolution of the dark matter in the in-
falling objects during each of these events, with the
parent galaxy represented by a time-dependent, an-
alytic potential consisting of bulge, disk and (spher-
ical) dark matter halo components.

Phase II: Describing the baryonic evolution and final lu-
m,inous structure — Star formation histories were
assigned to each accreted object via a simple pre-
scription. The parameters of the prescription were
normalized so that the final stellar and gas content
of satellites infalling today were similar to those of
the Local Group dwarfs. The stars thus generated
were "painted" on to the dark matter particles so
as to reproduce the trends in structural properties
of Local Group dwarfs.

Phase III: Following the chemical enrichment history
— The chemical enrichment associated with the
star formation within each object was followed us-
i ng a leaky-accreting-box chemical enrichment code

(Robertson et al. 20051. The parameters of the

chemical enrichment model were tuned to repro-
duce the mass-metallicity relation of Local Group
dwarfs and the low alpha-element ratios seen in
Milky Way satellites.

With all free parameters now fixed, the resulting stellar
halos were found to have total stellar content (~ lO^L©)
and density profile compar able to the Milky Way's halo

( Bullock fc Johnston[ 2005 ) and simil ar abundance pat -
terns to stars in the Milky Way's halo (Font et al. 2006a I

They also contained satellite systems with roughly the
same number of high-surface brightness companions as
the Milky Way (neglecting the recent and lower surface-
brightness discover ies by [Willman et al.[[2005a|b[ |Be-|
lokurov et al.|2006 and others) , and whose members had

similar structural parameters to the Milky Way's satel-
lites. Note that more recent work, using semi-analytic
models in combination with fully self-consistent cosmo-
logical simulations of the formation of Milky- Way-type
galaxi es has found very similar results ( De Lucia & Helmi
|||2008|

3. RESULTS I: CHARACTERISTICS OF DEBRIS FROM
INDIVIDUAL ACCRETION EVENTS

In this section we consider each accretion event individ-
ually in order to characterize how the spatial and veloc-
ity scales, morphology, surface brightness and abundance
patterns of debris depend on the accretion time, luminos-
ity and orbit of the progenitor satellite. We first discuss
debris from unbound systems before going on to look at
debris associated with still-bound systems (in §3.5)

3.1. Morphology

We visually inspected plots of the surface density of all
1515 of our accretion events from an internal perspective
(i.e. an AitofF projection of the material as viewed from
the parent galaxy center). Considering only the 1362
cases where no bound core of particles remained (deter-
mined from their mutual gravity and velocities relative to
each other) we found that debris could be broadly char-
acterized as belonging to a small set of morphological
classes. Figure [l] illustrates the three main types in the
classification scheme we adopted. (Note that the method
for creating this and all other images in the paper is de-
scribed in the appendix.)

"Mixed" morphologies (example in left-hand panels)
show a nearly uniform distribution of debris,
equally spread above and below the plane of the
parent galaxy, with the maximum exploration of
the latitude set by the initial orbital inclination.
The average line-of-sight velocity at a given point
is typically low, with no evidence of global veloc-
ity gradients. The local velocity dispersion is high.
The external view of this type of debris typically
shows a donut shape, indicating full-mixing of un-
bound material throughout the entire phase-space
region available to that orbit. Debris with this
morphology would contribute to the smooth back-
ground of the stellar halo rather than adding dis-
tinct substructures.

"Cloudy" morphologies (example in middle panels)
show a few (typically 2 or 3) distinct density
maxima, each a few tens of degrees across and
only loosely (if at all) connected to each other by
streams of intervening material. There are strong
velocity gradients across these "clouds", and they
can have low local velocity dispersions. The exter-
nal view reveals debris strung along eccentric orbits
with pericenters close to the parent galaxy center.
Plumes lead out from the central galaxy to shells
of material spreading out at the apocenters of the
orbits.

"Great circle" morphologies (example in right-hand
panels) show nearly uniform debris distributions
aligned with a single great circle on the sky (view-
ing the debris from a solar perspective leads to

some distortion of this great circle alignment, al-
though the continuity of debris is maintained).
There can be gradients in the line-of-sight velocity
along the great circle, but more gentle than those
seen in "cloudy" morphologies. The local velocity
dispersion can be low, or high if the debris has be-
come multiply- wrapped. The external view shows
debris spread evenly along only mildly eccentric or-
bits.

Figure [2] shows examples of events that were categorized
as "transitions" between our three main morphological
classes — mixed-clouds, cloudy-great-circlcs and mixed-
great-circles in the left, middle and right panels respec-
tively.

Figure [3] summarizes the physical interpretation of our
empirical scheme by plotting the accretion time vs orbital
eccentricity (as characterized by J/Jdrc ~ the ratio of or-
bital angular momentum to the angular momentum of a
circular orbit of the same energy) for each of our accre-
tion events, color coded with the assigned morphological
classification. The right-hand panel illustrates our gen-
eral conclusions: "mixed" morphologies arise from events
accreted more than 10 Gyrs ago that have had time to
fully mix along their orbits; "clouds" are from events ac-
creted less than 8 Gyrs ago on eccentric (low J/Jdrc) or-
bits; "great circles" are from events accreted 6-10 Gyrs
ago on near circular orbits (J/Jdrc >0.5); the transi-
tion types "mixed-clouds" and "mixed-great-circles" cor-
respond to events that are partially, but not yet fully
mixed along their orbits; and "cloudy-great-circle" mor-
phologies correspond to recent events on moderately ec-
centric orbits.

3.2. Surface brightness

Figured shows the maximum surface brightness of de-
bris associated with events of a given accretion time from
both external (upper panels) and internal (lower panels)
viewpoints. This maximum was selected by looking along
lines-of-sight to a random set of 10% of the luminous
particles in each simulation, and estimating the surface
brightness from the total luminosity and sky coverage
of their 30 nearest neighbors. (Using too few neighbors
gave noisier results, and using too many systematically
decreased the surface brightness estimates.) In this and
all subsequent plots, since we are interested in features
that are likely to stand out distinctly from the parent
galaxy's main components (i.e. be observable!), we only
consider material at greater than 30kpc projected sepa-
ration in the external view, or with distance moduli in
the range than 15-21 (~ 10-160 kpc) in the internal view.
For the same reason, we do not consider any events that
were classified as purely "mixed" morphology (transition
types are included).

The left-hand panels of Figure |4] are color-coded ac-
cording to the satellite luminosity at the time of ac-
cretion. As might be expected in a hierarchical forma-
tion context there are few high-luminosity events at early
times. Nevertheless, for a given accretion time there is a
clear trend for the higher luminosity events to result in
higher surface brightness-debris and in general the more
recent events have higher surface brightness (as a con-
sequence of the shorter phase-mi xing time — see |John-
stmll [19981 IHelmi fc W"hitel[T999l [Johnston, Sackett, fc

Fig. 1. — Plots to ill ustrate the general characteristics of mixed (left hand panels), shell/plumes/clouds (middle panels) and great circle
right hand panels) mor phologies, ibp panels show external views in surface brightness (with boxsizc scaled arbi trarily). Lower pane ls
slTo^^tTrsCyXironproJections of surface density, mean line-of-sight velocity and velocity dispersion.

Fig. 2. — As Figure |1| for transitions — mixed-clouds (left hand panels)
[right hand panels) — bcEwccn the three morphological classes

cloudy-great-circles (middle panels) and mixed-great-circle

12

^

l^

10

^\

9

reat

shells,

''

rcles

8

plumes \
or clouds\

/stil

6

bound

0.0 0.2 0.4 0.6 0.8 1.0

J/Jcl.

0.0 0.2 0.4 0.6 0.8 1.0

J/Jdrc

Fig. 3. — Age vs J/ Jcirci color coded with morphology: blue and
cyan indicate clouds and cloud/mixed morphologies; red and ma-
genta indicate great-circles and great-circle/mixed morphologies;
purple indicates cloud/great-circle morphologies; black indicates a
mixed distribution; and orange stars are for still-bound satellites.
The right-hand panel shows the dominant morphology contributing
in each portion of this plane.

■ . |. ijcrccl »i™ ^

■■?* " ■

. clouds/GlJWL

■ clouds .-'VL

■ -i«cl/clcu* l^h

6 8 10 12 14

t„„, (Gyrs)

(Gyrs)

Fig. 4. — Maximum surface brightness of debris associated with
accretion events in all eleven halos as a function of accretion time.
Upper panels are from an external perspective and lower panels
are from an internal perspective. Open diamonds are for debris
from still-bound satellites. Black/orange/yellow/blue/red points
in left-hand panels are for objects with luminosities < 10^/10^ —
lO'^/lO'' - 10^/10'^ - 10*/ > 1O**L0. Right-hand panels are color-
coded with morphology using the same scheme as Figure [3]

|Bullock||2001[ for more in-depth discussion of timescales
for phase- mixing of debris.) We conclude that the most
obvious debris features observed around galaxies today
should come from the most recent and most luminous
accretion events.

The right-hand panels of Figure [4] are the same as
the left-hand panels, but color-coded according to the
morphological type of the debris. As shown in p.l[
the transition mixed-clouds and mixed-great-circle types
arise from earlier accretion events and hence have lower
surface brightnesses. For events accreted at a given
time, those with great-circle morphologies tend to have
higher surface brightness. Nevertheless, the highest sur-
face brightness events overall are predominantly cloud or
cloudy-great-circle morphology. This is because our sam-
ple contains no events of great-circle morphology that
were disrupted within the last 6 Gyears — the progeni-
tors of such debris would be recently accreted satellites
on near-circular orbits and these are still bound in our
simulations.

Fig. 5. — Angular scales around the highest surface brightness
point in the debris, color coded with morphology using the same
scheme as in Figure [3| The red dotted line indicates the scaling
given by equation ([ill.

3.3. Phase-space scales

Previous studies of satellite disruption along mildly ec-
centric orbits show that the range in energy, angular mo-
mentum and orbital inclination in debris are a function
of the progenitor satellite mass mgat via the tidal scale

(msat/Mcal)

(1)

where Moai is the mass of the parent galaxy en closed
within the pericenter of th e satellite's orbit (e.g. John-

ston] [1998] [Warnick et~al] [2008). The range in orbital
energies corresponds to a range in orbital time periods,
and this leads to the debris spreading along the orbit to
form tidal streams. The range in angular momentum cor-
responds to a range in the precession rate of the "petals"
of the rosette orbits, and this leads to thickening of the
streams in the orbital plane. The range in orbital incli-
nation corresponds to a range in the precession rates of
the orbital plane, and leads to additional thickening of
the streams perpendicular to the orbital plane. The net
effect of all this spreading is that the streams become
less dense and (locally) colder over time (as noted in §1,
and seeJHelmi & White 1999). On the other hand, once
the streams are multiply-wrapped (i.e. have spread more
than ±7r in azimuthal phase along the orbit) it is possible
for an apparently local sample (i.e. selected spatially) to
contain debris from several distinct orbital phases and
have a large dispersion (e.g. see bottom panels of Figure

Figure [5| illustrates some of these ideas by plotting
the angular scale of debris as viewed from the center
of the parent galaxy for each of our accretion events,
color-coded by debris morphology (and again omitting
"mixed" morphologies). In cases of "great-circle" or
"great-circle/mixed" morphology (red or pink points)
the scale represents the width of the best-fit great cir-
cle containing 25% of the debris. The width of debris
with "great-circle" morphology roughly follows the ex-
pected scaling (i.e. ex s — see equation [l]) indicated
by the red line, even though variations in the mass-to-
light ratio of the different objects (due to differing masses
and accretion times) means that the translation between
mass and luminosity is not strictly linear. The pink
points ("cloudy-great-circles") fall systematically above
this line because these events have had longer to spread
beyond this width. If the precession of the orbital plane
precession is small enough, their large width could also
be due to multiple-wrapping in addition to the steady
spreading that goes along with mixing.

The blue points represent events with "cloudy" (dark
blue) and "cloudy-mixed" (cyan) morphologies. In these

-0.5
-1,0
-1,5
-2,0
-2,5
-3,0
-3,5
-4,0

IrV-

6 8 10 12 14
=„ (Gyrs)

(Gyrs)

Fig. 6. — Mean [Fe/H] (left hand panels) and [a/Fe] (right hand
panels) of debris associated with accretion events in all eleven ha-
los as a function of accretion time. Black/orange/yellow/blue/red
points in left-hand panels are for objects with luminosities <
lO^/lO^ - lO'^/lO'' - 10^/10^ - 10*/ > 10*Lq.

cases the angular scale represents the radius contain-
ing 25% of the light from the highest-surface-brightness
cloud. This radius clearly follows a different (and gener-
ally larger) scaling than the great-circle width, indicative
of a different mechanism for the origin of debris on ec-
centric rather than circular orbits (mass loss via sudden
shocking rather than steady stripping) . The trend for the
older debris (as indicated by the more mixed "transition"
type morphologies) to be more spread-out remains.

3.4. Stellar populations

Figure [6] plots the average [Fe/H] (left hand panel) and
[a/Fe] (right hand panel) as a function of the accretion
time for each satellite, with the points color coded ac-
cording to satellite luminosity. The main trends in the
figure can be understood by recalling our assumptions
that a satellite falling in today should sit on the observed
mass-metallicity relation for dwarfs (hence lower lumi-
nosity dwarfs at any epoch have lower [Fe/H] , as seen in
the left-hand panel), and that all star formation ceases
once a dwarf falls into the parent galaxy. The latter as-
sumption means that a satellite's accretion time sets the
length of its star formation history: those accreted long
ago will have short histories, with their abundance pat-
terns dominated by contributions from supernovae Type
n (hence high [a/Fe]); those accreted more recently will
have long histories, with their abundance patterns con-
taining by contributions from both supernovae Type H
and supernovae type lA (hence low [a/Fe], as seen in
the right -hand panel). These ide as are discussed in more

depth in Robertson et al. (2005).

Note that the chemical properties of recently accreted
material in our models (and substructure and surviving
satellites in particular) are very well constrained since
they depend on characteristics of model satellites that are
fixed by requiring agreement with observations of Local
Group dwarfs. However, the properties of earlier accre-
tion events are likely to be less accurately reproduced
since they depend on chemical evolution that is an ex-
trapolation of current observational constraints. In par-
ticular, while the general trends in abundance patterns
as a function of accretion time and satellite mass seen
in Figure [6] can be thought of as fairly robust, the abso-
lute values and scatter are indicative only and should be
treated with some caution.

3.5. Debris associated with surviving satellites

We deferred discussion of debris associated with still-
bound systems since these represent a particular subset

of events whose properties have been selected by dynam-
ical shaping: the stars in the left-hand panels of Figure
[3] show that surviving satellites mostly come from recent
events on mildly-eccentric orbits (as seen in Font et al.
|2006a ). In addition, of the 153 survivors in our sam-
ple, only 33 have lost more than 1% of their luminous
matter and only 13 have lost more than 10%, so debris
from these objects is not typically a major contributor
to stellar halos.

As might be expected from this summary, visual in-
spection of images to classify the morphology of debris
from survivors revealed no debris for the majority of
cases. Of the remainder 18 had debris localized around
the satellite, 22 had great-circle morphologies and 6 had
cloudy-great-circle morphologies. The open diamonds in
the upper panels if Figure |4] show the maximum surface
brightness of debris from still-bound satellites is gener-
ally lower surface brightness than that from completely
disrupted satellites that were accreted at the same time.
Nevertheless, in our sample of 11 halos there are six still-
bound satellites with associated debris brighter than 30th
mag/arcsec^.

Overall, these results suggest that while finding a single
satellite clearly in the process of disruption with debris
spread around the parent galaxy should not be surprising
in sufficiently sensitive surveys, finding many examples
of satellites in this state around a single galaxy would be
unusual. Moreover, since surviving satellites tend to be
on mildly-eccentric orbits we might expect debris asso-
ciated with Galactic satellites to mostly be along great
circles — indeed the debris streams from the disruption
of the Sagitt arius dwarf galaxy have just such a morphol-
ogy (see e.g. Majewski et al. |[2003 l.

4. RESULTS II: CHARACTERISTICS OF SUBSTRUCTURE
IN STELLAR HALOS

In this section, we build on the previous results to
examine how the characteristics of substructure in stel-
lar halos depend on the general properties of their con-
stituent satellites — i.e their merger histories. Note that
we only include contributions from unbound systems in
our analyses — although surviving satellites can con-
tribute debris we do not expect a significant fraction
of either the total content or number of substructures

within a stellar halo to come from survivors (see ^3.5 1
so this simplification should not affect our results

The results of j 3.1|3.3 already emphasize the funda-
mental limitation of using coordinate-based studies to
probe accretion histories back beyond the last few Gyears
to the epoch when the most rapid growth is likely to have
occurred for Milky- Way-type galaxies (i.e. > 8 Gyears
ago): older debris is fully mixed, morphologically indis-
tinct and low surface brightness, hence hard to attribute
to individual events and interpret in terms of accretion
histories. However, Figure |6] clearly demonstrates that
the early accretion epoch is precisely when we expect
the most rapid evolution in abundance patterns to oc-
cur and these can provide a complementary probe of the
histories at early times.

The black lines in Figure [7] illustrate the merger histo-
ries of our eleven "standard" halos by plotting the frac-
tional contribution of events to the total luminosity of the
halo as a function of accretion time, satellite luminosity
and satellite orbit. The bulk of these halos are built from

1,0

/ J^} I'i '

.08

\ M' \

E

\ 1 [f^ 1

i, D.6

/ / Jii^

f_

1 ( fm, '

\_ 0.4

IJM;

0,2

MM

0.0

^^^^"^^

1,0

'

'

1

f"^#

= 08

/ :

"^yyi

S °«

-f^H

H\ 'i

h"

/

'M

■f '' i

0,2

F

W:-

'-' /

0,0

t

6.0 6.5 7.0 7.5 8.0 8.5 9.0 0.0 0.2 OA 0.6 0,8 1.0

Fig. 7. — Fractional contribution of satellites to the total lumi-
nosity of the stellar halo as a function of accretion time, satellite
luminosity and orbit type. Black lines are for our eleven stan-
dard halos. Halos built from ancient/recent events are shown in
green/blue. Halos built from high/low luminosity events are shown
in yellow/orange. Halos built from events on radial/circular orbits
are shown in cyan/purple.

accretion events occurring more than 8 Gyears ago, with
luminosities greater than several times IO'^Lq and on a
mixture of radial and circular orbits.

In order to explore how sensitive the appearance of
substructure is to different merger histories, six "arti-
ficial" model halos (each containing roughly IO^Lq in
stars) were constructed by simply summing over accre-
tion events from our entire library chosen to have the
required properties:

• our "ancient" and "recent" halos (green and dark
blue lines in Figure nh were built from events en-
tirely accreted longer than 11 Gyears ago or more
recently than 8 Gyears ago;

• our "high luminosity" and "low luminosity" halos
(yellow and orange lines in Figure ul were built
from events either more or less luminous than sev-
eral times IO^Lq;

• our "radial" and "circular" halos were built from
events predominantly on radial or circular orbits
(with J/Jcirc < 0.2 or .J/Jdrc > 0.75 — cyan and
purple lines in Figure^.

Of course, while these histories clearly lie outside the
region in mass, orbit and accretion-time expected for the
majority of galaxies in a ACDM cosmology (as indicated
by the black lines) , it is not impossible that some galaxies
in the Universe were formed in just this way.

4.1. Number of features

We estimated the number of substructures that could
be seen around our model galaxies by counting all events
that were not of "mixed" morphology and contributed
more than 50% of the total stellar halo light at their
brightest point (i.e. to be considered above the back-
ground). Figure p] shows the cumulative number of satel-
lites contributingieatures brighter than /Xmax in sll eleven
"standard" halo models (black lines) as well as our six ar-
tificial halo models from both external (left hand panel)
and internal (right hand panel) viewpoints.

The black lines show that distinct debris features
brighter than 26th mag/arcsec^ should be unusual, while
surveys reaching to 30th/35th mag/arcsec^ can be ex-
pected to see of order a few to a dozen features around
Milky- Way type galaxies. The simplified nature of our
modeling is likely to prolong the clarity of substructure.

1 25
3 20

Mexternal .

^ \

i 15

\\ V u

1 '°

— ^V \\

""^x \

^ 5

''sfev.vy

=fe

:±;iji^.^

1 •="

c

° 25

S 20

_circ'uior orbiMi

high L°oi K
_ recent \\

'i 15

— . c\\ W

01

■^\ \ \\

3 10

^1^ \ \ w

VxVk, \ \\

^ 5

VO^^J^L Vv '

»

, \^^^-

40 38 36 34 32 30 28 26
^ (mag/orcsec^)

2 3 4 5

// (LOG(LQ/degree^))

Fig. 8. — Left hand panel: Number of accretion events with
debris above the background (see text) and brighter than fj, at pro-
jected distances greater than 30kpc from the host. Right hand
panel: the same thing from an internal perspective for debris
with distance moduli in the range 15-21. Black lines are for our
eleven standard halos. Halos built from recent events are shown
in blue. (Note that all events in our "ancient" halos resulted in
debris with diffuse morphology that was not counted in this fig-
ure.) Halos built from high/low luminosity events are shown in
yellow/orange. Halos built from events on radial/circular orbits
are shown in cyan/purple.

which suggests that the lines in Figure |8] should be con-
sidered upper limits on the number of satellites likely to
be contributing. However, our analysis considers only
the brightest point of each event's debris and in real-
ity debris from a satellite might contribute several dis-
tinct observable features. Hence these numbers are best
thought of as order-of-magnitude estimates for what we
might see. (Note that these numb ers are rather greater
than the lower limits estimated by [Johnston, Sackett, fc|
|Bullock|[2001[ in part because these authors only consid-
ered events that were accreted after the parent galaxy
had grown to 90% of its current size.)

The colored lines give us some idea of what this fre-
quency plot is telling us. The blue lines show that there
are many more features around our "recent" halo than
seen in our "standard" halos since debris from recent
events tends to have distinctive morphology and higher
surface brightness. Our "low luminosity" halo (orange
lines) has a comparable number of features to our "re-
cent" halo simply because of the sheer number of low
luminosity accretion events required to put it together
- it has a similar number of accretion events in the last
7 Gyrs as our "recent" halo even though the bulk of its
mass was accreted much earlier. The opposite is true for
our "high luminosity" halo (yellow lines). The number
of features around our "radial" and "circular" halos falls
within the distribution of our eleven "standard" halos be-
cause the accretion histories for these halos are close to
the standard ones. Finally, there are no green lines (cor-
responding to our "ancient" artificial halo) because the
satellites randomly selected to build it from had debris
all of "mixed" morphology — the lack of recent accretion
means that all events have had time to mix along their
orbits and leave no distinctive features. Overall, Figure
[8] suggests that the number of features around a galaxy
can be broadly related to the number of recent accretion
events.

4.2. Fraction of material in substructure

The amount of material in substructure is assessed in
Figure [9] by plotting the cumulative fraction of the stel-
lar halo contributed by the satellites with debris of non-
uniform morphology (i.e. counted in Figure pi) brighter
than surface-brigthness /i. The black lines show that we

a 1.0
?i 0.8
^ 0-6
ti 0.4

external

^

— \ \

c

-~- ^ "~VA

,8 0-0

T,,r,,'^.,-ri\,,>

40 38 36 34 32 30 28 26
/2 (nug/arcsec')

2 3 4 5

/i (LOG(LQ/degree^))

Fig. 9. — Fraction of halo luminosity contributed by objects
with associated debris brighter than fi and above the background
(as defined in the text). Black lines are for our eleven standard
halos. Halos built from recent events are showrn in blue. (Note
that all events in our "ancient" halos resulted in debris with diffuse
morphology that was not counted in this figure.) Halos built from
high/low luminosity events are shown in yellow/orange. Halos built
from events on radial/circular orbits are shown in cyan/purple.

/I (LOG(Lo/degree'))

/I (LOG(La/degree'))

Fig. 10. — Repeats right-hand panel of Figure Is] for features of
mixed/cloud, cloud or cloud/great-circle morphologies (left hand
panel) and great-circle or great-circle/mixed morphologies (right
hand panel). Black lines are for our eleven standard halos. Halos
built from recent events are shown in blue. (Note that all events
in our "ancient" halos resulted in debris with diffuse morphology
that was not counted in this figure.) Halos built from high/low
luminosity events are shown in yellow/orange. Halos built from
events on radial/circular orbits are shown in cyan/purple. No pur-
ple lines appear in the left-hand panel because no debris with cloud
morphology was found in the halo built from circular orbits.

should in general expect ^^10% of the stars in a stel-
lar halo to be in the form of distinct features, with a
large range (1-50%) around this typical value. Although
there are more features by number fainter than /i =30
mag/arcsec^ (see Figure p| , a larger fraction of the ma-
terial in substructures in the stellar halo are associated
with features brighter than this.

The colored lines in Figure [9] suggest that the primary
factor determining the amount of material in substruc-
ture is the epoch of accretion. The clear outliers in Fig-
ure [9] are the blue and green lines which correspond to
"recent" and "ancient" halos respectively, while those
with more usual accretion epochs fall within the range
of our "standard" models. In particular, note that even
though Figure Is] shows that there are similar numbers of
features in the "recent" and "low-luminosity" halos the
former contains 80-90% of its material in substructure,
while the latter has a more typical value of ~25%.

4.3. Morphology of features

Figure 10 repeats the right hand panel of Figure 8] (the
frequency of features in general) for debris classined as
having cloud (left hand panel ~ including mixed/cloud,
cloud and cloud/great-circle types) and great-circle mor-
phologies (right hand panel - including great-circle and
great-circle/mixed types). Note that there are roughly
equal numbers of cloudy and great-circle morphologies

1.0

lli^l^^-1

■p 0.8

5

-S 0.6
3

■5 0.4
c
.0

g 0.2
0.0

1

i : ,'

50 100 150

r (kpc)

Fig. 11. — Number fraction of maxima of cloud, mixed-cloud and
cloud/great-circle morphologies (left-hand panels) or great-circle
and great-circle/mixed morphologies (right hand panel) within dis-
tance r of Sun.

for features in our standard halos (comparing the black
lines in the left- and right-hand panels) and low- or high-
luminosity halos (orange and yellow lines). Our stellar
halos built from radially-biased/circularly-biased orbits
(cyan and purple lines respectively) show distinctive pat-
terns with a larger fraction of debris in cloud/great-circle
morphology . Hence the distribution of distinct debris
morphologies in halos should tell us something about the
distribution of orbital properties of the recently accreted
satellites.

Our "recent" halo also appears to contain more clouds
rather than great circle morphologies. This is a reflection
of the fact that surviving satellites tend to be relatively
recent accretion events in near circular orbits — debris
in the recent halo will tend to come from objects that
were completely destroyed and be biased towards radial
orbit events.

4.4. Radial distribution of substructure

Figure [TT] plots the number fraction of features with
maxima within distance r of the Sun, separately for fea-
tures with "cloud" (left hand panel) and "great circle"
(right hand panel) morphologies. Neither panel contains
any features within lOkpc of the Sun, most of the great
circles lie in the range 30-50kpc and the clouds occupy
a larger range in distance (20-lOOkpc). Hence we expect
stellar halos to be increasingly dominated by substruc-
ture as surveys explore larger and larger distances from
the host galaxy, and at the largest distances we antici-
pate that most of this substructure would be in the form
of clouds.

4.5. Angular scales of features

Figure [12] illustrates the distribution of angular scales
for debris of non-uniform morphology (i.e. plotted in Fig-
ure Is]). In general, all our halos show a range of scales,
from a few to ten degrees across. As might be antic-
ipated from our understanding that debris with great-
circle morphology tends to have smaller angular scales
than debris with cloud morphology (see Figure ^ the ar-
tificial halos built from circular orbits (purple Ime) have
smaller angular scales than those built from radial orbits
(cyan line). The halo built from low luminosity events
(orange line) also has smaller scales than the typical stan-
dard halo. Hence the typical scales of substructure re-
flects both the luminosity function and orbital distribu-
tion of infalling objects.

4.6. Stellar populations

10

Fig. 12. — Frequency of angular scales in debris, as seen from
an internal perspective. Black lines are for our eleven standard
halos. Halos built from recent events are shown in blue. (Note
that all events in our "ancient" halos resulted in debris with diffuse
morphology that was not counted in this figure.) Halos built from
high/low luminosity events are shown in yellow/orange. Halos built
from events on radial/circular orbits are shown in cyan/purple.

Fig. 13. — Luminosity-weighted average [a/Fc] against [Fe/H]
for entire content of each halo (solid points). The open points in-
dicate luminosity-weighted averages for the events that contribute
clear substructure (i.e surface brightness maxima above the back-
ground, as seen from an internal perspective).

Figure IlS] plots the average [a/Fe] against [Fe/H] for
the entirety of (soHd points) and substructure within
(open points) our stellar halos. In nearly every case,
the substructure is chemically distinct fr om the halo as a
whole. As noted in Font et al. ( 2006a|b ) this distinction
can be interpreted as corresponding to the distinction in
accretion epoch of the stellar halo in general (>8 Gyrs
ago) vs substructure (< 8 Gyrs ago). The one exception
to this trend is the "recent" stellar halo which was all
accreted < 8 Gyrs ago.

The colored points show how this intuition can be be
applied to interpreting abundance patterns in stellar ha-
los in general. The green/blue points (ancient/recent
halos) bracket the more general distribution of points in
[a/Fe], while the orange/yellow points (low/high lumi-
nosity halos) bracket the distribution in [Fe/H]. Hence,
in our model halos the [a/Fe] distribution is telling us
about the epoch of halo accretion, while the [Fe/H] dis-
tribution is sensitive to the luminosity function of ac-
creted satellites. Conceptually, we can imagine splitting
the [Fe/H]-[Q;/Fe] plane into four quadrants correspond-
ing to accretion histories dominated by ancient, high-
luminosity events (high [Fe/H] and [a/Fe]), ancient, low-
luminosity events (low [Fe/H] and [a/Fe]), recent, high-
luminosity events (high [Fe/H] and low [a/Fe]) and re-
cent, low-luminosity events (low [Fe/H] and [a/Fe]).

As noted in |3.4| the absolute values in the above re-
sults will depend on the assumptions in our models for
how chemical enrichment within each dwarf proceeds.
Nevertheless, the trends apparent in Figure [13] demon-
strate the potential power of using abundances as a probe
of early halo histories — precisely the epoch that cannot

be probed by debris morphology alone because it has had
time to become fully mixed.

5. SUMMARY: INTERPRETING HALO PROPERTIES IN
TERMS OF ACCRETION HISTORIES

Having examined how the characteristics of stellar ha-
los are related to the properties of their constituent satel-
lites in island 34]we now go on to synthesize and illustrate
these results using phase- and abundance- space maps of
all of our halos.

Figure [T4| summarizes our expectations for the general
properties of stellar halos built within a ACDM universe
through images of one of our "standard" halos. The left
hand column shows external views of surface brightness,
average [Fe/H], [a/Fe] and velocity, and velocity dis-
persion along the line-of-sight. The right-hand column
shows surface densities of stars as viewed form a point
8 kpc from the center of the galaxy in a set of shells of
increasing radius. The figure illustrates that, for typical
accretion histories of Milky- Way type galaxies, a stellar
halo will show: of order 10% of its content in substruc-
ture, with most of that substructure beyond ~20kpc from
its center; a handful of features brighter than 30th mag
arcsec"^ and dozens brighter than 35th; roughly equal
numbers of these features with cloud and great circle
morphology; clouds (or plumes and shells) occurring over
a larger range of galactocentric distances (10-lOOkpc)
than great circles (or rosettes — 30-50kpc); '^ 1 of its
satellites with tidal tails brighter than 30th mag/arcsec^;
and abundance patterns that are distinct for stars in the
smooth component (a- rich), highest surface-brightness
substructures (often metal rich and always a-poor) and
surviving satellite population (typically metal poor and
a-poor) of the halo.

As an example of how these typical properties arise,
Figure [l5] shows the contribution to the images in the
left-hand column of Figure 14 from satellites of differ-
ent accretion times. Panels in columns 1/2/3 show con-
tributions from all satellites accreted more than 12/8/4
Gyears ago. These columns illustrate our general intu-
ition that the inner halo is built early on (more than 8
Gyears ago), and that the debris from these early events
is now fully mixed (see Figure p| . The brightest sub-
structures dominate the outer halo and typically come
from more recent events (see Figure E| . Surviving satel-
lites typically fall in o nly a few Gyears ago (see Bul-
lock & Johnston 2005[ ) . It is precisely because of their
late infall that both satellites and dominant substruc-
tures are chemically distinct from the bulk of th e stellar
halo ( [Robertson et al.||2005[JFont et al.||2006a|b[ ). Satel-
lites and substructures are often chemically distinct from
each other because the former are dominated (by num-
ber) by low-luminosity (and hence low-metallicity) ob-
jects, while the brightest (i.e. first detected) examples
of the latter are often fr om high-luminosity (and hence
high- metallicity) objects ( Font et al.[|2008[ [Gilbert et"ar
20081).

5.1. General trends in halo properties with accretion
histories

In reality, there is a large range in halo properties
around the general description given above, which can
be interpreted as being due to the corresponding range
in accretion histories. The results of S|3] and S|4[ clearly

11

Observable property

Interpretation

Implication

fraction in
substructure

recent
accretions

high fraction => many recent events
low fraction => few recent events

scales in
substructure

luminosity function
(and orbit type)
of recent events

large => high luminosity events
small => low luminosity events

number of
features

number of
recent events

large => many events
small => few events

morphology
of substructure

orbit
distribution

clouds/plumes/shells => radial orbits
great circles => circular orbits

[Fe/H]

luminosity
function

metal-rich => high luminosity events
metal-poor => low luminosity events

[q/Fc]

accretion
epoch

o-rich ^ early accretion epoch
«-poor => late accretion epoch

TABLE 1
Summary of general trends for stellar halo interpretation

demonstrate that the frequency of substructure and the
fraction of a stellar halo in substructure around a galaxy
is sensitive to both the epoch when it accreted most of
its mass and the luminosity scales of the objects it ac-
creted, while the morphology of the substructure reflects
the orbit distribution of the progenitor satellites. Unfor-
tunately, substructure in coordinate space only remains
apparent in surface brightness and morphology for suf-
ficiently long to probe the more recent accretion history
of a galaxy in detail, corresponding to approximately the
last 10% of its mass growth in a ACDM universe. How-
ever, we expect the abundance patterns in stellar popu-
lations to evolve most rapidly at early times, and hence
offer a complementary probe of the early accretion his-
tory of a galaxy. Table [l] summarizes our understanding
of phase- and abundance-space signatures of accretion
histories developed so far.

The intuition outlined in the preceding paragraph and
summarized in Table [T] is illustrated in Figures |16| |17|

ure 19 shows two clear examples of outliers in "observa-

tions^ of our "standard" halo models (halos 4 and 6)

and|18| which repeat Figure 14 for our paired "ancient"

/ "recent" , "high luminosity" / "low luminosity" and "ra-
dial" / "circular" artificial halos. When compared to our
"standard" example in Figure [14] stars in the "ancient"
halo are concentrated towards its center with no obvi-
ous substructure, small changes in line-of-sight veloc-
ity across the halo, large velocity dispersion in its out-
skirts, shghtly depressed [Fe/H] and enhanced [a/Fe].
In contrast, stars in the "recent" halo are almost en-
tirely in substructure with no obvious smooth compo-
nent, larger extremes in line-of-sight velocity across the
halo, small velocity dispersion in its outskirts, slightly en-
hanced [Fe/H] and significantly depressed [a/Fe]. Both
the "high luminosity" and "low luminosity" halos show
some smooth component, but the former contains a few
broad, bright substructures while the latter is crossed by
many thin substructures. In the "high luminosity" halo
[Fe/H] is significantly enhanced, while it is depressed in
the "low luminosity" case. Finally, stars in the "radial"
halo are smoothly distributed with some substructure in
the form of shells, plumes or clouds while stars in the
"circular" halo show abundant substructure in the form
of rosettes or great circles.

5.2. Worked examples

Our "artificial" halos were deliberately constructed to
be outliers through the manipulation of their accretion
histories (as shown by the colored lines in Figure [?]). Fig-

which we can use, along with halo 8 (Figure 14 1 and Ta-
ble IT] to test to whether our intuitive interpretations are
accurate.

Conside r fir st the spatial and kinematic distributions
in Figure K^ (halo 8). There are several high surface
brightness features on a variety of spatial scales that indi-
cate recent accretion of both low and high-mass objects.
These recent events have not yet fully mixed, so there
are also large areas of low-dispersion in the outskirts of
the galaxy, with the average line-of-sight velocity varying
between extremes of ± 150 km/s. The mixture of cloud
and great circle morphologies suggest a range of orbital
properties in these recent events.

These interpretations are reinforced by the spatial dis-
tribution of the abundance patterns. In [Fe/H] halo 8
shows an average background of ^ —1.0, with clear dis-
tinct substructures superposed and offset by ~0.5 dex to
both higher and lower metallicity. These substructures
are uniformly lower in [a/Fe] (~ 0.05 — 0.1) than the
background {^ 0.2). The distinct morphology and low
[a/Fe] of these substructures suggest recent (< 8Gyrs
ago) accretion, with their variety of metallicities point-
ing to both low- and high-luminosity progenitors. The
background, fully mixed debris probes further back in
the galaxy's history than the distinct substructure. The
high [a/Fe] and intermediate [Fe/H] values of this mate-
rial are indicative of an epoch of early accretion (^ 10
Gyrs ago) of intermediate mass objects.

In contrast to halo 8, the surface brightness distribu-
tion of halo 4 (left hand panels of Figure [l9| is rela-
tively smooth, with only moderate (± 50 km/s) varia-
tions in line-of-sight velocity and high dispersion in the
outskirts of the galaxy, suggesting a well-mixed popula-
tion with little recent accretion. The lack of great-circle
morphologies points towards a lack of recent accretions
on circular orbits. The background of halo 4 has a higher
average metallicity and slightly lower [a/Fe] than halo 8.
The high background [Fe/H] suggests significant contri-
bution from just a few high-mass progenitors, while the
low [a/Fe] suggests that the main accretion epoch was
slightly later than usual.

Finally, the surface brightness plots for halo 6 (right-
hand panels of Figure 19 1 show many small-scale sub-

structures that are not yet fully- mixed, and tend to have

12

LL (maq/crcsec'^}
33. 55. 52. 29. 26. 25.

[Fe/H]

M (LOGCU/degreeO)
0. 1. 2. 4. 5, 6. 7.

M (LOG(U/degree^)}

[«/Fe]

'i„ (km/s)

0. 1. 2. 4. 5, 6. 7.
M (LOG(U/degree^)

D. 1. 2. 4. 5, B. 7.

M (LOG(Le/degree''»

-200. -100. 0. 100. 200.
C7 (km/s)

0. 1. 2. 4 5, 6. 7,
M (LOGCU/dsgree^))

0. 50. 100. T50.

D. 1. 2. 4 5. 6. 7.
1

Fig. 14. — Left hand column shows the surface brightness,
average [Fe/H] , [a/Fe] , Une-of-sight velocity and dispersion (panels
running top to bottom) for our standard halo number 8, as viewed
from an external perspective. Each panel is SOOkpc on a side.
Right hand column shows all sky projections (as viewed from a
point 8kpc form the center of the galaxy) of the surface density of
stars within different distance moduli ranges. The top right-hand
panel shows the full range of distance moduli considered (15-23, or
10-398 kpc)

great-circle morphologies. The outskirts of halo 6 has low
[Fe/H] and high [a/Fe] , while the inner parts have inter-
mediate [Fe/H] and [a/Fe] . The former should come from
a large number of low-luminosity, early accretion events,
while the latter indicates some intermediate mass and
age events.

Figure 20] which repeats Figure |7] with halos 4, 6 and
8 highlighted in color confirms our general conclusions:
~70% of halo 8 (shown in yellow) was accreted prior to 8
Gyrs ago, with a the remaining 30% in few large events
since then; ^60% of halo 4 (shown in red) comes from
a small set of accretion events of intermediate age (all
around 8 Gyrs ago) with less than 5% of the halo accreted
since that time and a bias towards radial orbits; and
^60% of halo 6 (shown in blue) comes from early (>10
Gyrs ago) accretions, with no large (> IQ^Lq) events
and most on circular orbits.

6. DISCUSSION: APPLICATION TO OBSERVATIONS

6.1. Current data

The Milky Way's stellar halo has been studied in detail
locally with solar neighborhood samples (in some cases
including fu ll phase- and/or abundance-space info rma-
tion, see e.g. Helmi, White, de Zeeuw, fc Zhao|19"99| and
more deeply (but typically only in small survey patches)
using brighter, photometric ally selected tracers (e.g. RR

Ly raes and giant stars — [Ivezic et al.|[2000{ [Vivas et
ar][200Tl [Morrison et~ar|l2000| [Majewski, Ostheimer,

Kunkel, & Patterson 2000). Truly global maps out to
distances of nearly lOOkpc have been made only in the
last few years using carbon stars and M-giants selected
from the 2 Micron All-Sky Survey (although limited to
a small number of tracers, rough distance estimates and
sparse line-of-sight velocity measu rements, see [Ibata et|

aLll^OOT: 'Majewski et al. 1 12003). Most recently, the

Sloan Digital Sky Survey (SDSS) has provided panora-
mas of one quarter of the sky as viewed with turnoff
star "glasses" and out to distances of ~ 60 kpc (Be-

' ' In coordinate space, these studies

of our Galaxy have revealed a smooth, metal-poor inner
(<10kpc) component to its Galactic halo and several sub-
structures at larger (10-30kpc) radii that as yet have no
confirmed progenitor. Of the substructures, one appears
to trace only a mildly eccentric orbit (the Mon oceros

ring, see Newberg et al. 2002

Ibata et al. 20031, while

there are three examples of dinuse clouds (in the constel
lations of Vir go, Triangulum- Andromeda and Hercules
Aquila — see[Juric et al.[[2008l [Rocha-Pinto et al.„2004j

Majewski et al.[[2004j [Belokurov et al.|[2007p . [Belokurov
et al. ] 2006| report the discovery of a grea t-cir c le aligned
strean i (see also |Grillmair & Dionatos 2006 Grillmair

2006 1 , which has tentative ly been associated wit h the

Ursa Major dwarf galaxy (Fellhauer et al. 20071, and
ther e are several more such streams from"gIobular clus-

ters (Odenkirchen et al.|2003 Grillmair & Johnson'2006|
Grillma ir fc Dionatos||2006| Two of the Milky Way's
other s atellites (of more than a dozen — see "Belo kurov]
et al.|2006) have material within a few tidal radii of their
main bodies with spatial and velocity signatures consis-
tent with tidal debris, although this interpretation i s nei-

ther u nique n or conclusive (L eo I and Carina, see Sohn
et al.[]20071 [Munoz et al.[]2008l ). Onl y one satellite (tEe^
Sagittarms dwarf galaxy, Sgr, see — [Ibata, Gilmore, fc]
Irwin 19941 has definitive evidence for tidal tails. Sgr's

13

Fig. 15. — Contribution to standard halo 8 in surface brightness, average [Fe/H], [o/Fe], line-of-sight velocity and dispersion (panels
running top to bottom) from all satellites accreted before 12/8/4 Gyrs ago (in columns from left to right). Each panel is SOOkpc on a side.

14

Li (mag/arc3ec^) fi (LOG(Le/deqree'^)) |

3S, 35. 32. 29. 2.6, 2J.

-Z.O -1.5 -1.0 -Q.5
[«/Fe]

■■■■'i™ (km/a)

0. 1, 2. 4. 5. 6. 7.
15<tJ^<23

M (LOG(Le/degree^)}

0. 1. 2. 4. 5, 6. 7.
Ji tUPSCLa/degres^))

0, 1. 2. 4. 5, 6. 7.
M (LDG(U/degree^))

-200. -100. 0. 100. 2oa
u (km/r,)

0. 1. 2. 4. 5, 6. 7.
M (LOG(L„/degree^)}

0. 50. IDO, 150

C. 1. 2. 4. 5. 5. ^.

V

,ii (mag/orcsec^) jti (LOG(Le/degree'')) |

38, 35. 32- 29. 26. 2J.

0. 1. Z .4- 5, 6. 7.
M (LOG(Ls/«Jegree^)

^1^

[Fe/H]

-2.0 -1.5 -1.0 -0.5

0. 1. 2. 4. 5. 6. 7.

n

i
1

[,ei/Fe]

M (LOSCLs/fegres'*))

-0.1 O.C- &.1 0.2

0. 1. 2. 4. 5. 6. 7.

J

' 1''-'l™i--lS

v« fttrn/»)

M (LOG(L^/degree^))

-200. -100. D.

K!i!SJ^^9l

0. 1. 2. 4. 5. 6. 7.

ITJP

1* ^

cr (km/s)

M (LOG(LE/degree"))

C. 50. 100. 150.

C. 1. 2. 4. 5. 6. ^.

Fig. 16. — As Figure [Til but for our
panels) accretion events.

'artificial" halos constructed from the most ancient (left hand panels) and most recent (right hand

15

it (maq/arcsec'^)
38, 35. 32. 29. 26. XJ.

M (LOG(L^/degreeO)

0. 1. z .*. 5, a. 7.

[Fe/H] ^ (LOG{LE/degree^»
-2.0 -1.S -1.0 — 3.& 0. 1. 2. 4. 5. 6. 7.

^'-^^^^^^^

0, 1. 2. 4. 5. B. 7.

M (LOGCL^/degree^))
0. 1. 2. 4. 5. 6. 7.

l&<d^.;21

fi (LOG(LE/dsgree^))

0. 1. 2. 4. 5. 6. 7.
21<;d^<23

Fig. 17. — As Figure
luminosity accretion evenl

but for our "artificial" halo constructed from the highest (left hand panels) and lowest (right hand panels)

16

Li (mag/arc3ec^} fi (LOG(Le/deqree'^)) |

38, 35. 32- 29. 2.6, 2J.

-z.a -13 -1.0 -0.5

[«/Fe]
-Q.I Q.O 0.1 0.2

'■■■'to= (km/a)

0. 1, 2. 4. 5. 6. 7.
1 5-:;d^<23

M (LOG(U/degree'))

G. 1. 2. 4. 5, 6. 7.
/J. (LOG(U/degree^)

0, 1. 2. 4. 5, 6. 7.
M (LOGCU/degree^))

-200. -100. 0. 100. 200.

f

0. 1. 2. 4. 3, 6. 7.
1 9<d„^<21

M (LOG(L„/degree^)}

G. SO. 100, ISO.

C. 1. 2. 4. 5. 5. ^.
2 1 <d^<Z3

Fig. 18. — As Figure |14| but for our "artificial" halo constructed from the accretion events on the most radial (left hand panels) and
most circular (right hand panels) orbits.

17

ix (mag/orc

sec-}

m(

LOG(

U/degreeO)

3S,

35. 3S. 29.

J

26.

Si

1.0

23.

I

-0.5

0. 1.

2.

1

4. 5, 6.

7_

23

LOG(

U/degree^))

-2.Q

-1.5

0. 1. 2. 4. 5, 5. 7.
M (LOG(L^yegree^)

"3

I»/Fe.]

i.1

0.2

-0.1

0.0 c

0. 1.

4

1 7<d„^-;

2.

1

4. 5, 6.

7

1

■

]

1

1

19

LOG(

U/degree^))

v,„ (km/

^^^_

^

-200.

-100. 0.

T'OE).

*(?,

0. 1.

2.

4. S, 5.

7.

19<:d,^-;:

1

21

LOG(

r

^ (km/s

^^^^

.^^H

00,

150.

C. 1.

2.

4. 5. 6.

7_

I^H

I

1

^^^^H

■■

•
1

i

21<d^<:

23

f

M (LOG{LE/degreeO)
0. 1. 2. 4 5. 6. 7.

l=<;d^--~23

M (LOG{LE/degree=))
0. 1. 2. 4 5. 6. 7.

'=Sd,^<17

M (LCGCLa/begree^)

D, 1. Z 4 5. 5. 7.
1 7<-.d,^< 19

M (LOG{L^/degree^))

-200. -100. 0. 100. 200. 0, 1. 2. 4 5, 6. 7.

c- (km/s)
0. m. 100. 150.

15<:d^<2i

M (LOGlL^/degree^))

0. 1. Z 4 5. 6. 7.
21<d^<23

Fig. 19. — Repeats Figure [14] for "standard" halos 4 and 6.

18

,.-^-^

1.0

' !|

r

-1

^

i"'"

B 0.6

e

■5

1 OA

ff

D.2

k , , .

0,0

12 10 8 6

6,0 6.5 7.0 7.5 8.0 8.5 9.0 0.0 0.2 OA 0.6 0.8

Fig. 20.— Repeats Figure!?]
histories for halos 4, 6 and 8 nif
respectively.

for our standard halos, with the
ghlighted in red, blue and yellow

debris stretches entirely around the galaxy, is relatively
meta l -rich and is aligne d with a great circle ([Ibata et al.

2001 Majewski et al.

2003, Law etal.|[200^

Studies ot Andromeda have often been limited to look
at only a tiny fraction of the stellar halo, for example
taking deep color-magnit ude diagrams using the Hub-
ble Sp ace Telescope (e.g. Ferguson et al. 2005 Brown
et al.||2007j , or selecting giant stars photometrically us-
mg narrow-b and filters for subs equent follow-up spec-
troscopy (e.g. Gilbert et al.|2006 l. The first globa l views

of the halo, firs t out to distances of ~ 50 kpc (Fergu

son et al

quarter o? the gal axy out to

2002|| and recently expanded to explore one
150kpc from its center

( [Ibata et al. [2007[ ) , were put together by plotting over-
densities in the number of red stars selected using broad-
band filters to have the right colors and magnitudes to
be giant stars associated with Andromeda. These stud-
ies have revealed that, like the Milky Way, Andromeda is
dominated by substructure beyond its inner 20kpc, most

notably by the giant stellar stream (Ibata et al. 20011

which has been mapped to > lOOkpc from its center, and
whose progenitor was likely a high- mass (10^M(7)) satel-
lite on a fairly radial orbit (.Guh athakurta et al.| |2006j
Font et al.|2006c[|Geehan et al.|2 006 Fardal et al. 2006).
i'he giant stream and some otner (possibly associated)
substructures within lO's of kpc of Andromeda's cen-
ter appear to be relatively metal rich (metallicity~ —0.7
Ferguson et al. 2005), while there is increasing evidence
for a smoother, more metal-poor ext ended component
at larger radii (metal l icity ^-1.3, se e

Chapman et al. 2006 Kalirai etal

far as 165kpc from its center (IGilbert et al

Irwi n et al.||2005
e xtend ing as

In summary, it is clear that the expectations for galax-
ies built within a ACDM Universe outlined in S|5] are
broadly consistent with our knowledge of the Milky Way
and Andromeda galaxies. Comparing specifically, we
have found that a stellar halo built entirely from ac-
cretion within a ACDM context should have: (i) of or-
der 10 % of its stars is substructure with of a do zen
streams brighter than 35th mag arcsec"^ (see f|4.l| and
il4j| as seem s to be the case f or both Milk;

eTia. 20081 and Andromeda (Ibata et alT] 20071; (n

ffiell

the brightest of these streams more m etal rich than the

smooth component of the halo (see j4.6 and Font et

al

2008 ) as noted for both galaxies by" Gilbert et al.
( |2008p ; (iii) only ~ 1 surviving satellite (Sgr) with ex-
tended, debris streams brighter than 30th mag arcsec"^.
typically on a mildly eccentric orbit (see ^3.5 1; (iv) de-

bris morphologies consistent with both circular and ec-
centric orbits — great-circles or rosettes for the former
(e.g. Monoceros, Sgr and the Orphan stream around the

Milky Way) and clouds, plumes or shells for the latter
(e.g. the Virgo, Triangulum-Andromeda and Hercules-
Aquila clouds around the Milky Way and the Giant Stel-
lar Stream around Andromeda).

Finally, note that the question of whether the smooth
components of stellar halos are contributed primarily
from stars accreted during earlier events (as we have as-
sumed) , or formed in situ within the primary dark matter
halo (either in an early disk that was subsequently de-
stroyed — see |Abadi et al.|2003[ — or an early, gas rich,
rapid accretion event) remains unresolved.

6.2. Future prospects

Current observations do not yet allow us to build global
surface density, velocity, me tallicit y and abundance maps
(i.e. comparable to Figures 14|T9) of any halo. However,
the summary of data in j j6.1| demonstrates that signif-
icant portions of both the Milky Way and Andromeda
have been surveyed in density and (in the latter case)
in metallicity using photometric indicators. Similar data
for oth er galaxies are gradually be coming available (e.g.

M33 - McConnachie et al. 2006), wit h the prospect of

moving beyond the Local Group (e.g. Seth et al. 20071

and building samples of dozens of galaxies on the horizon
In addition, several projects in the next decade promise
orders of magnitude increases in the sizes of data sets dis-
cussed here: the entire Milky Way halo will be probed
by repeated all-sky surveys of stars capable of detecting
(for example) RR Lyraes out to hundreds of kpc (e.g.
Panoramic Survey Telescope and Rapid Response Sys-
tem and the Large Synoptic Survey Telescope); plans
for next generation spectrographs (e.g. the Wide Field
Multi-Object Spectrograph) include taking spectra of
hundreds of thousands of Milky Way halo stars spread
over 10% of the sky with sufficient resolution to provide
estimates of [a/Fe] as well as [Fe/H], and with a deep
enough magnitude limit to sample the entire halo; the
same instrument could map Andromeda's giant stars in
position, velocity, [Fe/H] and [a/Fe].

Coincidentally, the inner halo (< 10-20kpc) will be dis-
sected in all six phase-space dimensions (e.g. via ESA's
GAIA mission) , with sufficient numbers of chemical ele-
ments measured to track r- and s-process sources as well
as [Fe/H] and [a/Fe] (i.e. using WFMOS). The poten-
tial of such studies has not been explicitly discussed in
this paper since the methods for building our simulated
halos are least accurate in this regime, and r- and s-
process chemical evolution is not included. However, pre-
vious work has already pointed the promise of such data
sets in phase- ( Hclmi & White 1999: |Helmi fc de Ze euw
20001 iKnebe et al. 2005; Pcnarrubia et al.||2006|

and

abundance- (Bland-Hawthorn & Freeman 2003: lenner

et al. 2006 1 space and our own results indicate the sen
sitivity of abundance patterns to accretion history even
in 2-dimensions.

Combining these data sets from the Local Group and
beyond: (i) the larger number of galaxies in our samples
will allow us to move beyond making broad consistency
checks (such as outlined in the previous section) to di-
rectly constraining the characteristics of and variation
among o6sen;e(i recent accretion histories of galaxies; (ii)
the increase in the sizes of data sets will increase our sen-
sitivity to the smallest progenitor objects and allow us to
examine the entire luminosity function of galactic build-

19

ing blocks; and (iii) the additional dimensions in the data
sets (phase- and abundance- space) will provide probes
back to the earliest accretion epochs when we expect the
bulk of most galaxies were being assembled.

7. CONCLUSIONS

In this paper we have explored the connection between
the merger history of a galaxy and the present day struc-
ture of its stellar halo in phase- and abundance-space.

We have found that studies of substructure in the
surface-density and/or density distribution in stellar ha-
los are sensitive to the recent (< 8 Gyears ago) merging
history of a galaxy. These substructures can tell us about
the mass, orbit and accretion time of progenitors, while
the proportion of mass in the smooth component reflects
the importance of early events (whether in the form of
direct accretions from minor or major mergers or in situ
formation from a heated disk or monolithic collapse).
For a Milky- Way-type galaxy, built within the context
of ACDM, we expect of order few to tens of percent of
the halo to be in the form of substructure, for those sub-
structures to be increasingly dominant at large radii and
for the inner halo to be relatively smooth. These expec-
tations are broadly consistent with the current data for
the Milky Way and Andromeda galaxies .

Our analysis was based on relating the results of a sub-
jective morphological classification of debris morphology
from individual accretion events to the properties of the
progenitor satellite. As our sample of galaxies and sub-
structures within galaxies grows, the challenging nature
of reaching very low surface brightness, and the multiple
dependencies of substructure characteristics mean that
interpretations of individual features in terms of progen-
itor properties may not always be unique. Hence, for
specific comparisons of data and observations, it makes
sense to recast our results in terms of statistical mea-
sures of substructure rather than individual interpreta-

tions. For example. Bell et al. (20081 report on a prelim-
inary analysis of the SDSS halo turnoff-star data which
quantifies the level of substructure via the dispersion in
counts around a smooth background. The same analysis
applied to the model stellar halos presented in this paper
suggests a degree of substructure similar to the observa-
tions. More generally, these statistics need to be designed
to be sensitive to differences in accretion history, armed
with the knowledge developed here that: (i) the epoch of
galaxy formation sets the percentage of the stellar halo
contained in phase-space substructure (and its average
[a/Fe]); (ii) the number and mass scale of recently ac-
creted objects set the number, angular scales (and mean
metallicities) of substructure; and (iii) the orbit type of
progenitors set the morphology of substructure.

In contrast to coordinate-space, signatures in
abundance-space are not subject to mixing over time
so should last inde finitely. This promises a wa y to look
further back time (Bland-Hawthorn & Freeman 2003).
Our own results demonstrate that variations in [a/i'e]
and [Fe/H] among the smooth components of different
stellar halos reflect variations is the dominant epoch of
accretion and masses of progenitor objects respectively.
Moving to higher dimensions in abundance space should
allow even more precise interpretations.

JSB's work was enabled by startup funds at UC Irvine.
KVJ, AF, SS and SNL were supported through NSF CA-
REER award AST-0133617. BER gratefully acknowl-
edges the support of a Spitzcr Fellowship through a
NASA grant administered by the Spitzer Science Cen-
ter.

The authors thank the Aspen Center for Physics for
hosting a Summer workshop in June 2006 which enabled
useful discussion in the group, and the referee for helpful
comments on the paper.

APPENDIX
GENERATING THE IMAGES FROM PARTICLE DATA

The i mages presented in j|5l were created using an adapted version of the multi-dimensional density estimation code
EnBiD (iSharma & Steinmet3|2006|).

For the external galaxy images, the smoothing length hi of an N-body particle with luminosity rrii was flrst calculated
using the k nearest neighbor scheme in 3d space, with k = 64. The surface density maps were then generated by
calculating the contributions from particles at each pixel point at position r in a 512x512 image using the scatter
formulation of smooth particle hydrodynamics (SPH) in two dimensions:

Pir) = ^ W{r - Ti; hi)m„

(Al)

where W is the kernel function and mi and r^ are the particle luminosity and position respectively. This estimator
gives the true density fleld in the sense that when applied globally the total mass is conserved. The average values of
velocities and abundances were estimated using a simple luminosity-weighted averaging:

A{v)

Y, W{v - Yi] hi)miAi
Y,W{v-v^]hi)m,

(A2)

where A is the average of quantity Ai .

For the Aitoff projections each N-body particle was split into multiple particles of equal mass and distributed in 3d
space using a kernel function and a smoothing length corresponding to 64 neighbors. The number of particles n that
an N-body particle is split into was set equal to the number of pixels the N-body particle encompasses in the Aitoff
projection. For an N-body particle of smoothing length /i at a distance r , n = ivh? / {Sr^) , S being the solid angle of
each pixel in the Aitoff image. The particles were then binned in an Aitoff projection to generate luminosity density
maps. The average values of velocities and abundances were calculated (as above) using a luminosity weighted average
over the number of particles in each bin.

20

REFERENCES

Abadi, M. G., Navarro, J. F., Steinmetz, M., & Eke, V. R. 2003,

ApJ, 597, 21
Bell, E. F., et al. 2008, ApJ, 680, 295
Belokurov, V., et al. 2006, ApJ, 642, L137
Belokurov, V., et al. 2007, ApJ, 657, L89
Bland-Hawthorn, J., k. Freeman, K. C. 2003, ASP Conf. Ser. 297:

Star Formation Through Time, 297, 457
Brown, T. M., et al. 2007, ApJ, 658, L95
Bullock, J. S., Kravtsov, A. V., &; Weinberg, D. H. 2001, ApJ, 548,

33
Bullock, J. S., & Johnston, K. V. 2005, ApJ, 635, 931
Chapman, S. C, Ibata, R., Lewis, G. F., Ferguson, A. M. N., Irwin,

M., McConnachie, A., & Tanvir, N. 2006, ApJ, 653, 255
Choi, P. I., Guhathakurta, P., & Johnston, K. V. 2002, AJ, 124,

310
De Lucia, G. & Helmi, A. 2008 astro-ph 0804.2465
Fardal, M. A., Babul, A., Geehan, J. J., &; Guhathakurta, P. 2006,

MNRAS, 366, 1012
Fellhauer, M., et al. 2007, MNRAS, 375, 1171
Fenner, Y., Gibson, B. K., Gallino, R., & Lugaro, M. 2006, ApJ,

646, 184
Ferguson, A. M. N., Johnson, R. A., Faria, D. C, Irwin, M. J.,

Ibata, R. A., Johnston, K. V., Lewis, G. F., & Tanvir, N. R.

2005, ApJ, 622, L109

Ferguson, A. M. N., Irwin, M. J., Ibata, R. A., Lewis, G. F., &

Tanvir, N. R. 2002, AJ, 124, 1452
Font, A. S., Johnston, K. V., Bullock, J. S., & Robertson, B. 2006,

ApJ, 638, 585
Font, A. S., Johnston, K. V., Bullock, J. S., k. Robertson, B. E.

2006, ApJ, 646, 886

Font, A. S., Johnston, K. V., Guhathakurta, P., Majewski, S. R.,

& Rich, R. M. 2006, AJ, 131, 1436
Font, A. S., Johnston, K. V., Ferguson, A. M. N., Bullock, J. S.,

Robertson, B. E., Tumlinson, J., & Guhathakurta, P. 2008, ApJ,

673, 215
Geehan, J. J., Fardal, M. A., Babul, A., & Guhathakurta, P. 2006,

MNRAS, 366, 996
Gilbert, K. M., et al. 2006, ApJ, 652, 1188
Gilbert, K. M., et al. 2008, in preparation
Grillmair, C. J. 2006, ApJ, 645, L37
Grillmair, C. J., & Dionatos, O. 2006, ApJ, 643, L17
Grillmair, C. J., & Dionatos, O. 2006, ApJ, 641, L37
Grillmair, C. J., & Johnson, R. 2006, ApJ, 639, L17
Guhathakurta, P., et al. 2006, AJ, 131, 2497
Helmi, A. & White, S. D. M. 1999, MNRAS, 307, 495
Helmi, A., White, S. D. M., de Zeeuw, P. T., & Zhao, H. 1999,

Nature, 402, 53
Helmi, A., & de Zeeuw, P. T. 2000, MNRAS, 319, 657
Helmi, A., White, S. D. M., & Springel, V. 2003, MNRAS, 339,

834
Helmi, A., Navarro, J. F., Nordstrom, B., Holmberg, J., Abadi,

M. G., & Steinmetz, M. 2006, MNRAS, 365, 1309
Ibata, R., Martin, N. F., Irwin, M., Chapman, S., Ferguson,

A. M. N., Lewis, G. F., & McConnachie, A. W. 2007, ApJ,

671, 1591
Ibata, R. A., Irwin, M. J., Lewis, G. F., Ferguson, A. M. N., &

Tanvir, N. 2003, MNRAS, 340, L21

Ibata, R., Irwin, M., Lewis, G., Ferguson, A. M. N., & Tanvir, N.

2001, Nature, 412, 49
Ibata, R. A., Gilmore, G., & Irwin, M. J. 1994, Nature, 370, 194
Irwin, M. J., Ferguson, A. M. N., Ibata, R. A., Lewis, G. F., &

Tanvir, N. R. 2005, ApJ, 628, L105
Ivezic, Z. et al. 2000, AJ, 120, 963
Johnston, K. V., Sackett, P. D., & Bullock, J. S. 2001, ApJ, 557,

137
Johnston, K. V. 1998, ApJ, 495, 297
Juric, M., et al. 2008, ApJ, 673, 864
Kalirai, J. S., et al. 2006, ApJ, 648, 389
Knebe, A., Gill, S. P. D., Kawata, D., & Gibson, B. K. 2005,

MNRAS, 357, L35
Law, D. R., Johnston, K. V., & Majewski, S. R. 2005, ApJ, 619,

807
Majewski, S. R., Ostheimer, J. C, Rocha-Pinto, H. J., Patterson,

R. J., Guhathakurta, P., & Reitzel, D. 2004, ApJ, 615, 738
Majewski, S. R., Skrutskie, M.F., Weinberg, M.D. & Ostheimer,

J.C. 2003, , ApJ, 599, 1082
Majewski, S. R., Ostheimer, J. C, Kunkel, W. E., & Patterson,

R. J. 2000, AJ, 120, 2550
Majewski, S. R., Munn, J. A., & Hawley, S. L. 1996, ApJ, 459, L73
McConnachie, A. W., Chapman, S. C, Ibata, R. A., Ferguson,

A. M. N., Irwin, M. J., Lewis, G. F., Tanvir, N. R., & Martin,

N. 2006, ApJ, 647, L25
Morrison, H. L., Mateo, M., Olszewski, E. W., Harding, P., Dohm-

Palmer, R. C, Freeman, K. C, Norris, J. E., & Morita, M. 2000,

AJ, 119, 2254
Morrison, H. L., et al. 2008, ArXiv e-prints, 804, arXiv:0804.2448]
Muiioz, R. R., Majewski, S. R., & Johnston, K. V. 2008, ApJ, 679,

346
Navarro, J. F., Helmi, A., & Freeman, K. C. 2004, ApJ, 601, L43
Newberg, H. J., et al. 2003, ApJ, 596, L191
Newberg, H. J. et al. 2002, ApJ, 569, 245
Odenkirchen, M., et al. 2003, AJ, 126, 2385
Penarrubia, J., Benson, A. J., Martinez-Delgado, D., & Rix, H. W.

2006, ApJ, 645, 240
Penarrubia, J., et al. 2005, ApJ, 626, 128
Renda, A., Gibson, B. K., Mouhcine, M., Ibata, R. A., Kawata, D.,

Flynn, C, & Brook, C. B. 2005, MNRAS, 363, L16
Robertson, B., Bullock, J. S., Font, A. S., Johnston, K. V., &

Hernquist, L. 2005, ApJ, 632, 872
Rocha-Pinto, H. J., Majewski, S. R., Skrutskie, M. F., Crane, J. D.,

& Patterson, R. J. 2004, ApJ, 615, 732
Seth, A. , De Jong, R., Dalcanton, J. et al 2007, preprint j astro-|

ph/0701704)
Sharma, S., & Steinmetz, M. 2006, MNRAS, 373, 1293
Sohn, S. T., et al. 2007, ApJ, 663, 960
Somerville, R. S., & Kolatt, T. S. 1999, MNRAS, 305, 1
Warnick, K., Knebe, A., & Power, C. 2008, MNRAS, 385, 1859
Wetterer, C. J. & McGraw, J. T. 1996, AJ, 112, 1046
Willman, B., et al. 2005, ApJ, 626, L85
Willman, B., et al. 2005, AJ, 129, 2692
Vivas, A. K., et al. 2001, ApJ, 554, L33